Removing 50Hz Power Line Interference from EEG Signals Using Adaptive Filtering

During electroencephalogram (EEG) signal acquisition, 50Hz power line interference represents a common challenge. This occurs when electrical system operating frequencies (50Hz or 60Hz) infiltrate biopotential signals through electromagnetic induction or equipment conduction, significantly compromising EEG data quality. Adaptive filtering technology provides an effective solution to address this issue.

The core principle of adaptive filtering involves dynamically adjusting filter parameters based on input signal characteristics to achieve optimal noise suppression. Unlike traditional fixed filters, adaptive filters can track real-time variations in power line interference, such as frequency fluctuations or amplitude changes, enabling more precise noise elimination. In code implementation, this typically involves creating an adaptive filter object with configurable parameters like step size and filter length.

In EEG signal processing, commonly used adaptive filtering algorithms include Least Mean Squares (LMS) and Recursive Least Squares (RLS). The LMS algorithm features computational simplicity, making it suitable for real-time processing applications. Its implementation typically requires initializing filter weights and updating them using the formula: w(n+1) = w(n) + μ·e(n)·x(n), where μ is the step size, e(n) the error signal, and x(n) the reference input. The RLS algorithm demonstrates faster convergence rates but demands higher computational resources. For both methods, the critical implementation aspect involves constructing a reference channel to model power line interference, then subtracting this noise component from the original EEG signal using the adaptive algorithm.

Notably, while removing power line interference, adaptive filtering must avoid damaging useful EEG signals. Therefore, parameter selection (such as step size and filter order) requires optimization based on specific scenarios. Code implementations often include parameter tuning loops to achieve optimal performance. Furthermore, combining adaptive filtering with other techniques like notch filters or Independent Component Analysis (ICA) can further enhance denoising effectiveness. For instance, developers might implement a hybrid approach where adaptive filtering handles dynamic interference while notch filters address fixed-frequency components.